DSM Generation

Hello everyone, I have generated two DSM from two Sentinel-1 SAR SLC pairs (17th and 29th November, 2018 and 16th and 28th January, 2019) .However the elevation range that I am getting from both this DSMs is entirely different from the elevation range obtained from SRTM DEM, Cartosat-1 DEM and Terrasar-X DEM (90 Meter). I know that there must be difference in elevation values between a DSM and DEM. However as the difference is very large therefore it is questioning the reliability of the DSMs. I have attached the picture of these DEMs along with the two DSM that I have created from Sentinel-1 data. I have also attached the methodology for DSM generation using the SNAP software.

Can somebody help me to interpret these results. Furthermore, how can I improve the accuracy of DSMs that I have generated from Sentinel-1 SAR data.

Steps for DSM Generation using Sentinel-1 SLC pairs

  1. Co-registering the pairs using Back Geocoding method and with the help of SRTM 3 ARC Second DEM.

  2. Interferogram Formation

  3. TOPSAR Deburst

  4. Phase Filtering using Goldstein Phase Filtering

  5. Phase Unwrapping using Snaphu Software

  6. Phase to Elevation Conversion

  7. Range Doppler Terrain Correction of the DSM

Good evening everyone, I have two questions.
1-I have finished a processed . With your expertise I would like to know if the treatment is good.
2- I would like to know with this data is it possible to generate a DSM? and how?

looks very noisy. Can you please share a screenshot of the interferogram (after filtering)?
Is there much vegetation in your study area?

Goldistein phase filtering.


it is in a semi-arid zone
the savannah west africa

doesn’t look too bad - but for a proper topography we need a bit more fringes. What is the perpendicular baseline of the image pair? You can check via the InSAR stack tool.

Also, selecting elevation as an additional output after terrain correction (at best with SRTM 1Sec), helps you comparing the elevations from Sentinel-1 (named elevation_VV) with the SRTM ones (named elevation), for example by calculating a difference band.

the difference in band between elevation sentinel 1 vv and SRTM will you give me a DSM ?

no, both are more or less surface models (slightly different penetration depths because of different wavelengths). But as SRTM was bistatic (two antennas at a time) its quality is superior to Sentinel-1 repeat pass approaches.

Have you seen this? DEM generation with Sentinel-1 - Workflow and challenges

Is this the tool that allows you to see the perpendicular baseline?insar stack

No, it is called InSAR stack:

here is the capture

the perpendicular baseline is only 30m. For good topographic mapping you need an image pair of at least 100m. This will give you more fringes to describe the topography.

You find some instructions on the selection of image pairs in the tutorial: DEM generation with Sentinel-1 - Workflow and challenges

For example, the ASF Baseline Tool allows you to identify an image pair with 12 days (or even 6) and a larger perpendicular baseline.